General-purpose chatbots can recite a striker’s career goals. They are much shakier on what happens when the starting lineup is announced ninety minutes before kickoff. Snapp is betting that gap is a company.
The San Francisco startup emerged from stealth with $4.6M from investors including a16z Speedrun, Dorm Room Fund, Gold House Ventures, House Fund and Valia Ventures. It is led by Shawn Tsao, who co-founded food delivery company Caviar, later sold to Square and then to DoorDash in a $410M deal.
Thirty live data feeds and a benchmark claim
Snapp runs on more than 30 live data pipelines and claims its models answer sports questions 44% more accurately than leading general-purpose systems, and up to 3.2 times faster. Those numbers come from company-run tests with no independent validation, and they measure sports tasks specifically rather than general reasoning.
The product comes in two halves. A consumer app spanning more than 100 leagues has passed 20,000 downloads and connects with ESPN, Yahoo Sports, Sleeper, Kalshi and Polymarket. An enterprise platform targets professional teams for scouting, recruiting and in-game decisions, with a pilot at UC Berkeley’s football programme.
The strategic question is durability. As OpenAI and Anthropic models gain fresher access to live data, the moat narrows to pipeline engineering and domain tuning. Snapp’s argument is that sports fans and coaches will pay for answered questions, not for a better model.
